2
Abstract
Since the first cases of COVID-19 were reported in late 2019, countries have attempted to limit
the spread of the disease using different methods. The purpose of our research is to identify the
effectiveness of some of these different methods. To do this, we examined the mandates and
restrictions issued by the state of Maryland. The data that we used includes: positive increase,
death increase, hospitalization increase, and recovered. We then compared this data to the timeline
of restrictions and mandates imposed by the Maryland government. Our results showed that there
is a significant difference in the rate of COVID cases between the successive months of March
through July (p<0.01), but no significant difference between the successive months of July through
October. Our data suggests that mandates and restrictions, alongside levels of enforcement and
community behavior patterns, have an impact on COVID-19 rates.
3
Introduction
Since the first reported cases of COVID-19 were reported in late 2019, doctors and scientists have
been brainstorming ways to limit the spread of the disease. On 23 January 2020, the central
government of China imposed a lockdown in Wuhan and other cities in Hubei in an effort to
quarantine the center of an outbreak of coronavirus disease 2019 (COVID-19); this action is
commonly referred to as the Wuhan lockdown. Once the disease spread to other countries,
worldwide social distancing and national quarantines ensued. As a result of this, we experienced
one of the quickest recessions in our nation’s history. Unemployment skyrocketed, businesses
closed, unessential travel ceased. However, how much of an impact have these precautionary
measures in limiting the spread of COVID?
Just as countries have adopted different strategies to combat COVID so has the United States.
Some states have had loose restrictions while others have taken aggressive steps in an attempt to
limit the transmission of the disease. These different strategies are necessary because of the
different population density of the states; what may work in Wyoming may not be the best
approach in New York City. The strategy that Maryland has taken has loosely coincided with
recommendations from the CDC and the current administration. Since the first COVID case in
Maryland, we can look at how these restrictions have impacted the spread of the disease.
The best source that we have found with the most comprehensive dataset of COVID tracking data
is from The COVID Tracking Project. Our group will analyse the most pertinent statistical COVID
related information in the dataset. We will then compare the timeline of restrictions and mandates
to the dataset. The goal of our data analysis is to determine the efficacy of government
mandates/restrictions to combat the spread of COVID-19 in Maryland.
The goal of our data analysis is to determine the efficacy of government mandated sanctions to
combat the spread of COVID-19 in Maryland. The data set that we will use is from The COVID
Tracking Project. This provides the number of COVID cases and number of COVID deaths in
Maryland, along with other relevant statistics. We will compare the prevalence of the disease
with the timeline of government mandated restrictions and advisories. We will then be able to see
how effective these restrictions are at limiting the spread of the disease.
4
We hypothesize that the restrictions are having an effect on the spread of COVID. However, we
do not know how big of an effect it is having. Since we do not know, we are very interested to
complete our research to find which restrictions/mandates are effective, and which are not. Since
the beginning of COVID, we have already learned that some precautions are effective. In March,
Dr. Fuaci said that masks were not recommended, later, the CDC stated that masks are strongly
recommended. Then many governors, including Maryland governor, Larry Hogan, made masks
mandatory. Our goal is to explore whether the different lockdown measures have had any impact
on the number of positive cases, hospitalizations, and/or deaths.
Null Hypothesis; the lockdown measures mandated in Maryland had no significant impact on the
spread or number of COVID-19 cases witnessed.
Alternative Hypothesis; the lockdown measures mandated in Maryland had a significant impact
on the spread or number of COVID-19 cases witnessed.
Hypothesis testing: from our analysis, we shall then determine if the rate of positive cases increase,
actually increased or decreased due to the mandated lockdown measures. Depending on our
findings, we shall then decide to retain the Null hypothesis or Alternative hypothesis and which to
reject.
Related Work
COVID-19 is a type of virus that is similar to the common cold. It was first discovered in humans
in the 1960s. Since then, it has been studied in several labs around the world to better understand
the virus, and further vaccine research.
Face coverings- Early in the COVID-19 pandemic, the WHO, the CDC and NIH’s Dr. Anthony
Fauci discouraged wearing masks as not useful for non-health care workers. Now they recommend
wearing cloth face coverings in public settings where other social distancing measures are hard to
do.
5
According to The Association of American Physicians and Surgeons, “Surgical masks are loose-
fitting devices that were designed to be worn by medical personnel to protect accidental
contamination of patient wounds, and to protect the wearer against splashes or sprays of bodily
fluids. They aren’t effective at blocking particles smaller than 100 μm”. This is a problem because
COVID is approximately 0.12 microns. They were designed to protect against droplets, not
aerosols.
Are people wearing masks correctly? In a study in Singapore, data was collected from 714 men
and women. Of all ages, only 90 participants (12.6%) passed the visual mask fit test. About 75%
performed strap placement incorrectly, 61% left a “visible gap between the mask and skin,” and
about 60% didn’t tighten the nose-clip. Masks, which are already not very effective against
spreading COVID, are nearly useless if not worn properly.
Social distancing- How effective is social distancing and where did the “6 foot” rule come from?
Large respiratory droplets (>5 μm) remain in the air for only a short time and travel only short
distances, generally <1 meter. They fall to the ground quickly. This idea guides the CDC’s advice
to maintain at least a 6-foot distance. Larger particles land on nearby surfaces quickly, anything
over 500-micron will be airborne less than 1 second.
How far does COVID-19 travel in air? According to Renown Health Products, “A 100-micron
particle will fall for about 6.7 seconds in still air can travel about 6ft with no assistance from
additional airflow. Particles less than 100-micron have the potential to be entrained in the airflow
pathway for some time and particles <50-micron may not settle, dependent on particle size and
indoor airflow. Particles smaller than 7-micron are easily entrained in the airflow pathway, rather
than settling to the ground”. They can stay suspend in air and replication-competent for extended
periods, therefore, social distancing will not mitigate risk from particles this size. This data
suggests that close contact with a symptomatic person, even while wearing a surgical mask is not
an effective barrier. However, simply passing a person with symptomatic COVID will not put you
at a high risk of transmission.
6
Many different organizations have tracked COVID in countries around the world. They have put
the data into line graphed so we can visualize trends. One organization that has done this is
WorldMeter. From this, we are able to visually compare the countries’ COVID rates and tell if
their mandates and restrictions are working. For the most part, they all follow a similar trend- peak
in early spring, drop during the summer, and a second wave in the fall/winter months.
Methodology
The data used in this study came from the COVID Tracking Project, which is a volunteer
organization launched from The Atlantic and dedicated to collecting and publishing data tracking
COVID-19 outbreak throughout the United States. Data on COVID-19 testing and patient
outcomes from all 50 states, 5 territories, and the District of Columbia were collected on a daily
basis. Most of the data compiled were taken directly from the websites of local or state/territory
public health authorities. In a case where data were missing from these websites, the missing
information was supplemented with available numbers from official press conferences with
governors or public health authorities. The website contains data from March 14 to date. For the
purpose of this study, we limited our scope to Maryland, by examining reported data from March
14 through October 20. The dataset contained the following columns:
date, state, positive, positiveIncrease, positiveCasesViral, negative, negativeTestsViral, pending,
positiveTestsViral, totalTestsPeopleViral, totalTestsViral, totalTestEncountersViral,
negativeTestsPeopleAntibody, negativeTestsAntibody, positiveTestsPeopleAntibody,
positiveTestsAntibody, positiveTestsPeopleAntigen, positiveTestsAntigen,
totalTestsPeopleAntigen, totalTestsAntigen, hospitalizedCumulative, inIcuCumulative,
onVentilatorCumulative, hospitalizedIncrease, death, deathConfirmed, deathProbble,
deathIncrease, recovered, dataQualityGrade.
See appendix B for more description of each variable. Furthermore, for the purpose of this study,
we considered only certain variables that were found important after performing some basic data
wrangling (cleansing) as seen in the results.
We used R-Studio (Version 1.3.1073) as the integrated development environment for statistical
computation, analysis, and visualization of our data.
7
The reference periods we tried for estimating the effects of lockdown mandastes was available
data prior to March 16, 2020 when we had the first mandate that called for public schools to close.
Also, the time frame between each mandate served as some reference to preceding the preceding
executive mandates that were signed since many mandates were ordered at different times. We
examined how effects change over the following dates :
March 16, 2020: Schools closed, executive order closes public places
March 23, 2020: Hogan ordered nonessential businesses to close
March 30, 2020: Governor issued stay-home order
April 15, 2020: Hogan signed face-mask order
May 13, 2020: Hogan announced Stage One of reopening (effective 15 May)
July 29, 2020: Hogan expanded mask order, issues out-of-state travel advisory
Aug. 27, 2020: All schools authorized to reopen, Hogan says
Sept. 1, 2020: Maryland entered Stage Three of recovery plan.
Sept 21: The governor announced restaurants could expand indoor capacity from 50% to 75%
beginning at 5 p.m.
Statistical Analysis
We referenced the stratified sampling model, to examine whether statewide mandates actually had
an impact on the spread of COVID-19 in the state of Maryland. This method allowed us to divide
the total population-data into smaller groups or strata (months / periods related to mandate dates)
to better understand existing relationships and trends between different sampling groups (months,
weeks or days) and compare the pre-post mandate changes of COVID-19 spread in Maryland
over time. By using a stratified method, we were able to examine the timeline of restrictions and
mandates, and if the imposed mandates had an impact on the spread of COVID-19 in Maryland.
After we categorized into months, we plotted regression lines to see if months had different trends.
We then analyzed the data using one-way ANOVA and subsequent to finding statistical
significance we used a Tukey Test to find out which of the months were statistically significant
from one another.
Results
8
a) We looked into the number of positive cases viral from March through October. Positive cases
viral is defined as the total number of unique people with a positive PCR or other approved nucleic
acid amplification test (NAAT), as reported by the state or territory. This is equivalent to a
confirmed case as per the Council of State and Territorial Epidemiologists (CSTE) case
definitions.
The above graph shows a bar chart of Positive cases viral over a period of March to mid October.
As shown on the positive cases viral graph, Maryland appears to have a worse trajectory, showing
an average increasing trend with multiple spikes in COVID-19 cases.
However, Maryland has started restricting businesses, closing schools, and promoting social
distancing a bit earlier than others relatively. Lockdowns don’t cover the entire state. There’s a
patchwork of different restrictions by city and county. So, the numbers are trending upwards and
the graph shows an increasing trend.
b). Another interesting information that was worth investigating was the case fatality ratio. The
case fatality ratio is the proportion of individuals diagnosed with a disease and ended up dying
from that disease. Thus, a measure of severity among detected cases.
9
According to WHO, the Case Fatality ratio can be calculated by the below formula for an ongoing
pandemic.
Case fatality ratio (CFR) =
(Number of deaths from disease/(Number of deaths from disease + Number of recovered
from disease))*100
Although the case fatality is not a biological constant, nevertheless it is useful to represent the
magnitude of the disease at a given time, in a given population in a specific context.
The above scatter plot shows the case fatality ratio for COVID-19 in Maryland over time, 20
October 2020.
From the above graph, interpretation is that the CFR was much higher in the earliest stages of the
outbreak. But in the weeks that followed, the CFR declined, reaching as low for patients who first
showed symptoms.
This is because, according to the WHO, "the standard of care has evolved over the course of the
outbreak."
10
CFR can decrease or increase over time as responses change and differ according to the infected
population's location and its characteristics such as age or gender. Older people, for example, are
projected to see more COVID-19 CFR than younger populations.
Strict lockdown strategies together with a wide diagnostic PCR testing of the population were
correlated with a relevant decline of the case fatality ratio.
c). Deaths confirmed and hospitalized totals were graphed separately. These graphs tracked the
total deaths among confirmed COVID-19 cases and hospitalized cases in Maryland over time. The
data points on the deaths confirmed graph represent the cumulative total number of deaths reported
to public health by the date along the bottom.This graph presents data by the date a death was
reported as being associated with COVID-19.
Hospitalization graph is a key indicator for understanding the severity of this disease and the
pandemic’s impacts on the health care system.
The above graphs show a grid scatter plot of deaths confirmed and hospitalized against months.
Based on death data, the trend remains declining, and remains above the epidemic threshold. This
number, of course, will always rise, but will also – eventually – plateau. A cumulative total can
never fall.
The steep upward moving line on the hospitalized graph indicates an increase in the number of
hospitalised cases. The steeper the slope, the faster the total is increasing. Since the outbreak,
overall weekly hospitalization rates have increased.
11
d). A basic visualization of the total number of Marylanders who contacted COVID-19 and the
number who died was deemed necessary for the layman’s understanding.
This pie chart illustrates that 4% of Marylanders who contracted COVID-19 died. Another way to
discuss is the death-to-case ratio which is the number of COVID-19 deaths divided by the number
of COVID-19 cases within a certain time interval - our dataset is from March 5, 2020 to October
20, 2020. We calculated this to be 38.7. However, the purpose of data visualization is to
communicate information as clearly as possible. A death-to-case ratio might be meaningful to an
epidemiologist, but this simple pie chart can communicate the severity of COVID-19 much more
effectively in layman’s terms. If the entire Maryland population (6,046,000) were to become
infected, 241,840 Marylanders would die.
12
The above 2 pie charts contrast the differences in COVID-19 severity. While only 15% of patients
are hospitalized (we are assuming that most deaths were hospitalized), once a patient enters the
hospital their survival rate dramatically decreases. One of the most important aspects of lockdown
measures is their ability to ameliorate the limitations of the state’s hospital capacity. This pie chart
uses the cumulative numbers over the spring, summer, and fall to illustrate the recovery rate. The
new discoveries that are being made regarding the management and treatment of novel COVID-
19 probably impact the recovery rate, and thus are a major confounder to making associations
between lockdown measures and recovery rates over time.
While our ultimate goal is to do a time series analysis of the trends in positive cases,
hospitalization, recoveries, and deaths to examine whether the lockdown orders had a measurable
effect in Maryland. We used the dataset element “positiveIncrease'' instead of the “positive”
because the latter is cumulative while the positive increase was a unique daily number. The box
plot shows the month of May with a record high in the number of positive increases and March
with the least. April and July seem alike with similar numbers of cases recorded, and June, August,
September and October all share similar ranges of positive increase cases.
13
We are using the boxplot to brainstorm and to find clues. We know, for example, that there is a
lag between when the orders are issued and when we can expect to see changes in the number of
positive cases, yet we do not know how exactly long it takes for the changes to make an impact.
On March 30th the governor issued a stay-at-home order, and on April 15th the face mask order
was issued. We see a drop in numbers in June after 2 months of major shutdowns. The highest
daily cases (around 1200) is in May. Yet, it could possibly be due to an increase in testing capacity.
Stage 1 of reopening begins on May 13th, and we see the numbers rise again in July approximately
6 weeks after. The outlier in July could be as a result of the 4th of July celebrations. Schools
reopened at the end of August, but the medians (around 500 cases a day) have stayed fairly close
together thus far. We will examine these trends in further detail in our next deliverable.
Visualizing COVID-19 patterns lets people understand how and where the pandemic progresses
more concretely. Many states are now gradually experimenting by relaxing standards to see
whether COVID-19 can be evaluated.. This should give us an indication of whether the spread is
easing or worsening. We get daily COVID-19 updates with lots of data, figures, and graphs to see
if we flatten the curve with respect to the number of the new cases. The majority of these are based
only on the total number of new cases confirmed or the daily number. This does not include ample
data as to whether the situation improves, stabilizes or deteriorates. That's why we have to take
into account the number of people tested daily for COVID-19 .
14
Above is the scatterplot with loess regression line to show the trend of positive increase in cases
against months. According to our dataset , PositiveIncrease is the daily increase in positive, which
measures Cases (confirmed plus probable) calculated based on the previous day’s value. We used
the dataset element “positive Increase'' instead of the “positive” because the latter is cumulative
while the positive increase was a unique daily number. If we observe the graph initially there is a
striking upward trend, then started to decline this is because of initial phase 1 of reopening then
again it started to incline due to reopening of schools and partial reopening of business.
15
Above is the scatterplot with loess regression line to show the trend of death increase in cases
against months. According to our dataset, deathIncrease is Daily increase in death, calculated
from the previous day’s value. COVID-19 death rates are truly dropping as of mid-October. The
elevation angle of the daily deaths curve gradually shifted from upward to downward pressure
and tended to level out. The number of deaths caused by COVID-19 is one key metric that is often
referred to, but as with other COVID metrics, it is a challenge to measure accurately. The issues
involved in measuring COVID-19 deaths and argue that the change in the number of directly
observed COVID-19 deaths is the most reliable and timely approach when using deaths to judge
the trajectory of the pandemic in the Maryland, which is critical given the current inconsistencies
in testing and limitations of hospitalization data.
16
Above is the scatterplot with loess regression line to show the trend of hospitalized increase in
cases against months. According to our dataset, hospitalizedIncrease Is Daily increase in
hospitalizedCumulative, calculated from the previous day’s value. Hospitalization graph is a key
indicator for understanding the severity of this disease and the pandemic’s impacts on the health
care system. On a chart of hospitalized, look at how steeply the line is moving upward. The steeper
the slope, the faster the total is increasing. Since the outbreak, overall weekly hospitalization rates
have increased to an extent and then started to decline and maintained plateau as of mid-October
data.
Without intervention steps, every 3-4 days is an exponential doubling of events. The number of
cases in Maryland, unrestricted, would increase exponentially in the initial months. This would
result in 1660 cases by the end of March, and 21742 cases by the end of April, if there were 3 cases
on March 5. With this in mind, local authorities are taking drastic action. In order to deprive the
virus of additional human targets, other authorities are enforcing preventive measures such as
closing down schools and restricting social gatherings. As of May 19, Maryland had the most cases
confirmed on a regular basis.
17
We grouped each month into a separate data frame and plotted line graph to see if months had
different trends. Here the line graph shows the new daily cases from mid-March until October.
Each month's trend is depicted with different color on the graph for easier analysis. In the daily
number of confirmed cases we see high jumps and large fluctuations going back and forth. From
the daily new cases data, it looks like there is a strongly decreasing trend in the number of
confirmed cases in June.
18
As a part of stratified sampling method, we have decided to group dates into 5 periods for better
understanding of trends.
Period1 indicates timeline from declaring a state of emergency to issuing stay-home order.
Period2 indicates timeline from issuing stay-home order to stage one of reopening.
Period3 indicates timeline from stage one of reopening to expansion of mask order.
Period4 indicates timeline from expansion of mask order to stage 3 recovery plan.
Period5 indicates timeline from Stage 3 recovery plan to expansion of restaurants indoor capacity.
The graph shown above is grid chart showing scatterplots with regression line for Period1, Period2,
Period3, Period4, and Period5
19
Period1 graph shows a sudden spike in trend as in this time period government mandates are in
the initial stage. Period2 graph shows fluctuations but gradually increasing trend in positive cases.
period3 graph shows a declining trend to some extent and then increasing this is due to the stage
one of reopening. Soon after reopening cases came to control then suddenly a number of cases
increased. Period4 graph shows declining trend throughout and plateau for some period. Period5
graph shows an increasing trend because of school reopening , positive cases in children increased
this may be the potential reason and then declined due to stage 3 of the recovery plan.
Grouped bar chart of sum of each positives, hospitalization, and deaths by month
The above bar chart looks closer at the chosen categories, or months. We see that hospitalizations
and deaths mimic positive cases rates. Positive cases are more likely to be impacted by mandates
while hospitalizations and deaths are more likely to be related by hospital capacity and advancing
medical knowledge about COVID-19.
20
Grouped bar chart of mean and standard deviation of positive cases by month
Thus, we chose to look closer at the means and standard deviations of the positive cases. However,
we have not found any notable observations regarding each month’s data variance. The next step
after visualization is to commence a statistical approach.
f) Analysis of Variance
We ran one-way ANOVA to test if there is statistical significance between months. However,
before, we tested the data if it met the ANOVA model. We examined whether the assumptions of
the data were upheld.
i) Assumption: Normal Distribution of Each Group
Below is a sample of the plots. We see some outliers in May.
21
For May:
For August:
22
For September:
For October:
ii) Assumption: Homogeneity of Variances
To test the assumption of homogeneity of variance in our dataset, we used the Bartlett Test
which was statistically significant (p<0.01), that is this assumption appears to be upheld.
23
Finally, we ran the ANOVA model and found that there was a statistical significance difference.
The confidence interval we’ve chosen is 99%. The p-value is 2.2 x 10
-6
i.e. than <0.01. Since
this was statistically significant, a Tukey Test was used to conduct a post-hoc test in order to
further examine the difference between the months. The R results follow:
24
We noted that May, the highest months, is statistically significant from all other months. June,
the lowest month, is statistically significantly different from all other months, except for
September and October. We chose to focus on successive months and summarized the results in
the following table:
25
There is statistical significance between successive months from March until June. While a full
shutdown was recommended, the mandates in April and May were still fairly conservative.
However, the lockdowns eventually resulted in a June downward dip that was statistically
significant. In July, the cases spiked again. From our previous, line graphs we can see the peak
number of cases in July is mid-July which is 10 months after July 4. From July to October, the
shifts in positive cases are not significant. The data suggests the plateau is a result of no major
mandate orders were made during this period, thus the numbers stabilized.
26
Discussion and Limitations
The literature used event study analysis which enables them to test whether a single event, such as
a mandate enforcement date, has a significant impact. A time series analysis seemed more
appropriate but our data did not meet the assumptions of this model. Chiefly, it did have
stationarity; the mean, variance, and autocorrelation did not stay consistent over time. Instead, we
decided to take inspiration from the stratified sampling model and categorized time by month.
Grouping the data into various months, made it convenient to manage given that our data set was
incremental on a monthly basis.
The p-values may have been unusually low due to the fact that not all one-way ANOVA
assumptions were met. Our interpretation has been conservative due to the fact that we did not use
the most appropriate model for our data and hypotheses.
Moreover, we were not able to take into account some of the major contributing factors to COVID-
19 infection rates like the enforcement of mandates and how well the population adhered to the
social distancing portions of the mandate.
Our data was also not further stratified by age, health status, profession, or geographic area which
may have facilitated a more narrow analysis of the impact of COVID-19.
In conclusion, our data suggests that the statistical difference between months can be correlated
with mandates as well as public behavior.
27
Appendix A
R code available from:
https://drive.google.com/file/d/1p7Ri0x-NQt16qi0uASajXjJK7eY-ets1/view?usp=sharing
Dataset used available from:
https://drive.google.com/file/d/1hIfuWDDOr4QD3QrZDdR6RGmWD8mFuU8M/view?usp=shar
ing
Appendix B
Cases
Cases (confirmed plus probable) - API field name: positive
Total number of confirmed plus probable cases of COVID-19 reported.
New cases - API field name: positiveIncrease
The daily increase in API field positive, which measures Cases (confirmed plus probable)
calculated based on the previous day’s value.
Probable Cases - API field name: probableCases
Total number of probable cases of COVID-19 as reported by the state. A probable case is someone
who tests positive via antigen without a positive PCR or other approved nucleic acid amplification
test (NAAT), someone with clinical and epidemiological evidence of COVID-19 infection with
no confirmatory laboratory testing performed for SARS-CoV-2, or someone with COVID-19
listed on their death certificate with no confirmatory laboratory testing performed for SARS-CoV-
2.
PCR tests
Confirmed Cases or Positive PCR tests (people) - API field name: positiveCasesViral
Total number of unique people with a positive PCR or other approved nucleic acid amplification
test (NAAT), as reported by the state or territory. This is equivalent to a confirmed case.
Negative PCR tests (people) - API field name: negative
Total number of unique people with a completed PCR test that returns negative.
Negative PCR tests (specimens) - API field name: negativeTestsViral
28
Total number of completed PCR tests (or specimens tested) that return negative as reported by the
state.
Pending - API field name: pending
Total number of viral tests that have not been completed as reported by the state.
Positive PCR tests (specimens) - API field name: positiveTestsViral
Total number of completed PCR tests (or specimens tested) that return positive as reported by the
state.
Total PCR tests (people ) - API field name: totalTestsPeopleViral
Total number of unique people tested at least once via PCR testing, as reported by the state. The
count for this metric is incremented up only the first time an individual person is tested and their
result is reported. Future tests of the same person will not be added to this count.
Total PCR tests (specimens) - API field name: totalTestsViral
Total number of PCR tests (or specimens tested) as reported by the state or territory. The count for
this metric is incremented up each time a specimen is tested and the result is reported.
Total PCR tests (test encounters) - API field name: totalTestEncountersViral
Total number of people tested per day via PCR testing as reported by the state. The count for this
metric is incremented up by one for each day on which an individual person is tested, no matter
how many specimens are collected from that person on that day. If an individual person is tested
twice a day on three different days, this count will increment up by three.
Antibody tests
Negative antibody tests (people) -API field name: negativeTestsPeopleAntibody
The total number of unique people with completed antibody tests that return negative as reported
by the state.
Negative antibody tests (specimens) - API field name: negativeTestsAntibody
The total number of completed antibody tests that return negative as reported by the state.
Positive antibody tests (people) - API field name: positiveTestsPeopleAntibody
The total number of unique people with completed antibody tests that return positive as reported
by the state .
Positive antibody tests (specimens) - API field name: positiveTestsAntibody
Total number of completed antibody tests that return positive as reported by the state.
29
Total antibody tests (people) - API field name: totalTestsPeopleAntibody
The total number of unique people who have been tested at least once via antibody testing as
reported by the state.
Total antibody tests (specimens) - API field name: totalTestsAntibody
Total number of completed antibody tests as reported by the state.
Antigen tests
Positive antigen tests (people) - API field name: positiveTestsPeopleAntigen
Total number of unique people with a completed antigen test that returned positive as reported by
the state.
Positive antigen tests (specimens) - API field name: positiveTestsAntigen
Total number of completed antigen tests that return positive as reported by the state.
Total antigen tests (people) - API field name: totalTestsPeopleAntigen
Total number of unique people who have been tested at least once via antigen testing, as reported
by the state.
Total antigen tests (specimens) - API field name: totalTestsAntigen
Total number of completed antigen tests, as reported by the state.
Hospitalization
Cumulative hospitalized/Ever hospitalized - API field name: hospitalizedCumulative
Total number of individuals who have ever been hospitalized with COVID-19.
Cumulative in ICU/Ever in ICU - API field name: inIcuCumulative
Total number of individuals who have ever been hospitalized in the Intensive Care Unit with
COVID-19.
Cumulative on ventilator/Ever on ventilator - API field name: onVentilatorCumulative
Total number of individuals who have ever been hospitalized under advanced ventilation with
COVID-19.
Currently hospitalized/Now hospitalized - API field name: hospitalizedCurrently
Individuals who are currently hospitalized with COVID-19.
Currently in ICU/Now in ICU - API field name: inIcuCurrently
Individuals who are currently hospitalized in the Intensive Care Unit with COVID-19.
30
Currently on ventilator/Now on ventilator - API field name: onVentilatorCurrently
Individuals who are currently hospitalized under advanced ventilation with COVID-19.
New total hospitalizations - API field name: hospitalizedIncrease.
Daily increase in hospitalizedCumulative, calculated from the previous day’s value.
Outcomes
Deaths (confirmed and probable) - API field name: death.
Total fatalities with confirmed OR probable COVID-19 case diagnosis.
Deaths (confirmed) - API field name: deathConfirmed.
Total fatalities with confirmed COVID-19 case diagnosis
Deaths (probable) - API field name: deathProbable.
Total fatalities with probable COVID-19 case diagnosis
New deaths - API field name: deathIncrease.
Daily increase in death, calculated from the previous day’s value.
Recovered - API field name: recovered.
Total number of people that are identified as recovered from COVID-19.
Date - API field name: date
Date on which data was collected by The COVID Tracking Project.
References
31
1. CDC. (2020, August 7). Coronavirus Disease 2019 (C)VID-19). Centers for Disease
control and Prevention.
https://www.cdc.gov/coronavirus/2019-ncov/prevent-getting-sick/cloth-face-cover-
guidance.html
2. Coronavirus disease 2019. (2020, March 3). Wikipedia.
https://en.wikipedia.org/wiki/Coronavirus_disease_2019
3. Dr. Fauci Sets the Record Straight About Mask. (n.d). Www. Msn.Com. Retrieved on
October 27, 2020. From https://www.msn.com/en-us/health/medical/dr-fauci-sets-the-
record-straight-about-masks/ar-BB19Aof3
4. Estimating mortality from COVID-19. (n.d.). Www.Who.Int. https://www.who.int/news-
room/commentaries/detail/estimating-mortality-from-covid-19
5. Hypothesis Testing - Examples and case studies. (n.d).
https://www2.stat.duke.edu/courses/Fall11/sta10/STA10lecture21.pdf
6. Jacobs P, Ohinmaa AP. The enforcement of statewide mask wearing mandates to prevent
COVID-19 in the US: an overview. F1000Res. 2020;9:1100.
